Could not include topos theory - contents
(in category theory/type theory/computer science)
of all homotopy types
of (-1)-truncated types/h-propositions
basic constructions:
strong axioms
further
A universe in a topos is a topos-theoretic version of the notion of Grothendieck universe; see that page for general motivation and applications.
To free the notion from membership-based set theory, we must replace sets of sets by families of sets, just as in passing from power sets to power objects we must replace sets of subsets by families of subsets.
A universe in a topos $\mathcal{E}$ is a morphism $el\colon E \to U$ satisfying the axioms to follow. We think of $el\colon E \to U$ as an $U$-indexed family of objects/sets (fibers of $el$ being those objects), and we define a morphism $a\colon A \to I$ (regarded as an $I$-indexed family of objects) to be $U$-small if there exists a morphism $f\colon I \to U$ and a pullback square
The arrow $f$ is sometimes called the name of $a\colon A \to I$, since in the case when $I=*$, the arrow $f:* \to U$ points at the term in the universe $U$ representing the object $A$. (See also this discussion and references there.)
Note that $f$ is not, in general, unique: a universe can contain many isomorphic sets. With this definition, the pullback of a $U$-small morphism is automatically again $U$-small. We say that an object $X$ is $U$-small if $X\to 1$ is $U$-small.
The axioms which must be satisfied are:
Every monomorphism is $U$-small.
The composite of $U$-small morphisms is $U$-small.
If $f\colon A \to I$ and $g\colon B \to A$ are $U$-small, then so is the dependent product $\Pi_f g$ (where $\Pi_f$ is the right adjoint to $f^*\colon \mathcal{E}/I \to \mathcal{E}/A$).
The subobject classifier $\Omega$ is $U$-small.
Note that since $0\to \Omega$ is a monomorphism, (1) and (4) imply that the initial object $0$ is $U$-small. A predicative universe is a morphism $el\colon E \to U$ where instead of (4) we assume merely that $0$ is $U$-small; this makes sense in any locally cartesian closed category. In a topos, the generic subobject $1\to \Omega$ is a predicative universe, and of course a morphism is $\Omega$-small if and only if it is a monomorphism.
If we assume only (1)–(3), then the identity morphism $1_0\colon 0 \to 0$ of the initial object would be a universe, for which it itself is the only $U$-small morphism. On the other hand, if $\mathcal{E}$ has a natural numbers object $N$, we may additionally assume that $N$ is $U$-small, to ensure that all universes contain “infinite” sets.
Note that any object isomorphic to a $U$-small object is $U$-small; thus in the language of Grothendieck universes this notion of smallness corresponds to essential smallness. Roughly, we may say that (1) corresponds to transitivity of a Grothendieck universe, (3) and (4) correspond to closure under power sets, and (2) corresponds to closure under indexed unions.
We spell out in detail some implications of these axioms for the case that the topos in question is the Categeory of Sets according to ETCS, to be denoted $SET$.
Write $*$ for the terminal object in $SET$, the singleton set. Notice that for each ordinary element $u \in U$, i.e. $* \stackrel{u}{\to} U$, there is the set $E_u$ over $u$, defined as the pullback
We think of $E$ as being the disjoint union over $U$ of the $E_u$. In the language of indexed categories, this is precisely the case: the object $E\in SET$ is the indexed coproduct of the $U$-indexed family $(E\to U) \in SET/U$.
By the definition of $U$-smallness and the notation just introduced, an object $S$ in $SET$, regarded as a $*$-indexed family $S \to *$, is $U$-small precisely if it is isomorphic to one of the $E_u$.
If $S$ is a $U$-small set by the above and if $S_0 \hookrightarrow S$ is a monomorphism so that $S_0$ is a subset of $S$, it follows from 1) and 2) that the comoposite $(S_0 \hookrightarrow S \to *) = (S_0 \to *)$ is $U$-small, hence that $S_0$ is $U$-small. So: a subset of a $U$-small set is $U$-small.
Let $S$, $T$ and $K$ be objects of $SET$, regarded as $*$-indexed families $f\colon S \to *$, $T \to *$ and $K \to *$. Notice that $(SET\downarrow S)(f^* K, f^* T) \simeq (SET\downarrow S)(\array{K \times S \\ \downarrow^{p_2} \\ S}, \array{T \times S \\ \downarrow^{p_2} \\ S})$ is canonically isomorphic to $SET(K \times S, T)$. Since $\Pi_f$ is defined to be the right adjoint to $f^*\colon SET \to SET \downarrow S$ it follows that $\Pi_f f^* T \simeq T^S$ is the function set of functions from $S$ to $T$. By 3), if $S$, $T$ are $U$-small then so is the function set $T^S$.
Let $I$ be a $U$-small set, in that $I \to *$ is $U$-small, and let $S \to I$ be $U$-small, to be thought of as an $I$-indexed family of $U$-small sets $S_i$, where $S_i$ is the pullback $\array{ S_i &\to& S \\ \downarrow && \downarrow \\ * &\stackrel{i}{\to}& I }$, so that $S$ is the disjoint union of the $S_i$: $S = sqcup_{i \in I} S_i$. By axiom 2) the composite morphism $(S \to I \to *) = (S \to *)$ is $U$-small, hence $S$ is a $U$-small set, hence the $I$-indexed union of $U$-small sets $\sqcup_{i \in I} S_i$ is $U$-small.
By standard constructions in set theory from these properties the following further closure properties of the universe $U$ follow.
In summary
the sets $\emptyset, *, \mathbf{2}$ are $U$-small;
a subset of a $U$-small set is $U$-small;
the power set $P(S)$ of any $U$-small set is $U$-small;
the function set $T^S$ for any two $U$-small sets is $U$-small;
the union of a $U$-small family of $U$-small sets is $U$-small.
the product of a $U$-small family of $U$-small sets is $U$-small.
Just as ZFC and other material set theories may be augmented with axioms guaranteeing the existence of Grothendieck universes, so may ETCS and other structural set theories be augmented with axioms guaranteeing the existence of universes in the above sense. For example, the counterpart of Grothendieck’s axiom
would be
One can show, from the above axioms, that the $U$-small morphisms are closed under finite coproducts and under quotient objects. See the reference below.
Recall that an $\mathcal{E}$-indexed category is a pseudofunctor $\mathcal{E}^{op}\to \Cat$. The fundamental $\mathcal{E}$-indexed category is the self-indexing $\mathbb{E}$ of $\mathcal{E}$, which takes $I\in \mathcal{E}$ to the slice category $\mathbb{E}^I = \mathcal{E}/I$ and $x\colon I \to J$ to the base change functor $x^*$.
An internal full subcategory of $\mathcal{E}$ is a full sub-indexed category $\mathbb{F}$ of $\mathbb{E}$ (that is, a collection of full subcategories $\mathbb{F}^I\subset \mathbb{E}^I$ closed under reindexing) such that there exists a generic $\mathbb{F}$-morphism, i.e. a morphism $el\colon E \to U$ in $\mathbb{F}^U$ such that for any $a\colon A \to I$ in $\mathbb{F}^I$, we have $a \cong f^*(el)$ for some $f\colon I \to U$. In this case (since $\mathcal{E}$ is locally cartesian closed) there exists an internal category $U_1 \;\rightrightarrows\; U$ in $\mathcal{E}$ such that $\mathbb{F}$ is equivalent, as an indexed category, to the indexed category represented by $U_1 \;\rightrightarrows\; U$.
An internal full subcategory is an internal full subtopos if each $\mathbb{F}$ is a logical subtopos of $\mathbb{E}$ (closed under finite limits, exponentials, and containing the subobject classifier). A universe in $\mathcal{E}$, as defined above, can then be identified with an internal full subtopos satisfying the additional axiom that $U$-small morphisms are closed under composition.
In a topos with a universe, we can talk about small objects in the internal logic by instead talking about elements of $U$. We can then rephrase the axioms of a universe in the internal logic to look more like the usual axioms for a Grothendieck universe, with the morphism $el\colon E \to U$ interpreted as a “family of objects” $(S_u)_{u\colon U}$:
In a well-pointed topos, such as a model of ETCS, these “internal” axioms are equivalent to their “external” versions that refer to global elements of $U$.
Mike: I haven’t actually checked anything in this section, but it seems probable.
The construction of universes in presheaf toposes was first discussed in the following short note which might serve as a concise introduction to the more general reference that follows:
The general case is considered here: